Nothing. Earlier this month, Ross McKitrick from Canada’s University of Guelph and I published a manuscript in the Journal of Geophysical Research‐Atmospheres saying precisely that.
Scientists have known for years that temperature records can be contaminated by so‐called “urban warming,” which results from the fact that long‐term temperature histories tend to have originated at points of commerce. The bricks, buildings, and pavement of cities retain the heat of the day and impede the flow of ventilating winds.
For example, downtown Washington is warmer than nearby (and more rural) Dulles Airport. As government and services expand down the Dulles Access road, it, too, is beginning to warm compared to more rural sites to the west.
Adjusting data for this effect, or using only rural stations, the United Nations’ Intergovernmental Panel on Climate Change states with confidence that less than 10% of the observed warming in long‐term climate histories is due to urbanization.
That’s a wonderful hypothesis, and Ross and I decided to test it.
We noted that other types of bias must still be affecting historical climate records. What about the quality of a national network and the competence of the observers? Other factors include movement or closing of weather stations and modification of local land surfaces, such as replacing a forest with a cornfield.
Many of these are socioeconomic, so we built a computer model that included both regional climatic factors, such as latitude, as well as socioeconomic indicators like GDP and applied it to the IPCC’s temperature history.
Weather equipment is very high‐maintenance. The standard temperature shelter is painted white. If the paint wears or discolors, the shelter absorbs more of the sun’s heat and the thermometer inside will read artificially high. But keeping temperature stations well painted probably isn’t the highest priority in a poor country.
IPCC divides the world into latitude‐longitude boxes, and for each of these we supplied information on GDP, literacy, amount of missing data (a measure of quality), population change, economic growth and change in coal consumption (the more there is, the cooler the area).
Guess what. Almost all the socioeconomic variables were important. We found the data were of highest quality in North America and that they were very contaminated in Africa and South America. Overall, we found that the socioeconomic biases “likely add up to a net warming bias at the global level that may explain as much as half the observed land‐based warming trend.”